The Drum Awards for Marketing - Entry Deadline

-d -h -min -sec

Agencies Agency Models Digital Transformation

Will generative AI reignite demand for digital transformation?


By Sam Bradley | Senior Reporter

September 6, 2023 | 12 min read

Agency leaders are hopeful that the gen AI big bang will lead to renewed demand for transformation services. But is the tech actually there? And will it change the minds of cautious clients?

People sat around a desk during a presentation

What would gen AI-enabled digital transformation projects actually involve? / Unsplash

Digital transformation briefs usually carry a large price tag. That has made it an attractive segment for agency groups fighting to persuade client-side execs of the value of traditional advertising and the effectiveness of their media-buying. And during the pandemic, when many companies were reviewing their business models, that additional income helped speed the global holding company recovery. As of August, Japanese advertising group Dentsu takes 33% of its worldwide revenue from the segment.

But those big price tags made transformation projects prime targets for spending caution once the UK and US economies began slowing again in 2022. And although inflation has begun to ease, enthusiasm in boardrooms for transformation projects hasn’t come back into fashion. At S4 Capital, owner of Media.Monks, a July trading update revealed that income from transformation clients had fallen. Executive chairman Sir Martin Sorrell credited the decline to “longer sales cycles” – in plain English, clients hesitating to invest while they wait for the global economy to recover.

Stagwell, the US-based holding company behind agencies such as Crispin Porter + Bogusky, Assembly and Forsman & Bodenfors, recorded a decline in transformation revenues. In response, president Mark Penn predicted that generative AI would provoke new demand for that category.

He told The Drum: “There will be new business transformation work. Just as every client originally needed websites and then every client needed a mobile app, now they’re going to need generative AI experiences. That will create a large layer of work.”

What might that actually look like? And can generative AI tools add value to transformation services today, or is this just another handwave statement?

Are gen AI-led transformation projects feasible?

Accenture Song’s James Temple, who is global lead of metaverse at the consulting giant, says the company has already found several applications for gen AI in this line of work. It used an OpenAI (the company that produces ChatGPT) generative AI text app, combined with other software from Microsoft, to automate responses from hotel managers to customers at Radisson hotels. By cutting the time managers had to spend responding individually to customers, the solution freed them up to deal with “more high-value challenges” in the day-to-day business of running a hotel.

Similarly, it’s begun automating the text product detail pages, using bespoke large language models or LLMs (or, at least, modified versions of GPT) on behalf of “a couple of large-scale e-commerce partners,” says Temple.

The idea is to reduce the amount of time dedicated previously to writing lengthy, country-specific copy for e-commerce webpages. “In the old model, there would have been a team authoring the original and then going through a huge localization process,” he explains. Now, that process has been sped up significantly.

Temple expects these examples to pile up in the coming months. A survey of 600 CMOs conducted by Accenture for its own research found that 56% expect gen AI to create stronger customer experience design. Temple also claims that 33% of CMOs are already making investments in gen AI. “We’re seeing… significant demand in our inquiries right now,” he says.

At Merkle, global chief strategy officer Matt Naeger says that it’s been able to deploy chatbots using generative AI to overhaul one client’s customer service approach. “We built a system behind their contact response platform and they were able to answer a massive number of incremental, inbound requests, compared with what they were able to do with a call center,” he says. The business was able to field more requests than before and reserve its call center capacity for the complex, time-consuming queries that actually required a human.

Crucially, Naeger says the project employed a large knowledge model (LKM), in conjunction with software using the more familiar LLM. In essence, it’s a training dataset but not from a linguistic corpus – it is one containing the relevant assets and information specific to a company.

Currently, Merkle is using off-the-shelf LLMs for these kinds of projects through its partnerships with Google, Microsoft and Adobe, rather than developing its own – the cost of which, Naeger notes, is “just massive.” Cheaper to develop, bespoke LKMs are “a big differentiator” for it, he adds.

‘The way they want the internet to work’

Marcos Angelides, chief strategy and innovation officer at Publicis Groupe company Spark Foundry UK, predicts that CMOs will be forced to invest and reimagine their UX and CX approaches by changing consumer demand.

“We’re seeing a lot of categories starting to be impacted,” he says. “Travel is a big one. The ability for people to create customized itineraries from ChatGPT – if you’ve got a bespoke need from a destination then you might spend a lot of time trying to create that yourself.” Embedding a chatbot or program that relied upon a generative AI text application, he argues could allow an airline, booking platform or hotel chain to provide a more involved discovery and purchase experience to prospective customers.

“As people adopt generative AI, that’s the way they’re starting to want the internet to work. But websites aren’t set up for that,” he says. As more advertisers and brands realize the need for an overhaul, he argues, they’ll turn to agencies.

Suggested newsletters for you

Daily Briefing


Catch up on the most important stories of the day, curated by our editorial team.

Ads of the Week


See the best ads of the last week - all in one place.

The Drum Insider

Once a month

Learn how to pitch to our editors and get published on The Drum.

Convenience and customer design aren’t the only considerations CMOs will be keeping in mind. Wunderman Thompson’s Josh Loebner says that accessibility should be a major consideration when integrating or building new experiences around generative AI – as well as an opportunity for agencies.

The firm’s global head of inclusive design, Loebner tells The Drum that gen AI tools are a potential solution to some accessibility woes. He points to Microsoft Teams’ live caption feature, which incorporates AI technology. “Something as simple as captions and transcripts in Teams could be a fulcrum for change,” he says. But the failings of gen AI tools – in particular, sample bias in training datasets and the potential biases of the people developing them – must also be kept in mind.

“The arbiters of artificial intelligence, I think more often than not, are individuals and teams building out AI platforms who tend to be non-disabled,” Loebner says. “We tend to think of accessibility as a project that we start and complete. But when we think about digital transformation, I would argue accessibility is an ongoing, cumulative and methodical process.”

To navigate those ongoing concerns – and avoid encoding technological biases into their UX design, brands will need guidance. Loebner says careful application of AI can help here, too.

The agency has deployed gen AI tools to speed up the process of auditing a brand’s digital presence for accessibility concerns. He says: “Wunderman Thompson incorporates Brand Guardian as a tool that our clients and internal teams use regularly and repeatedly within digital transformations… to ensure assets within digital platforms are accessible and inclusive. It makes things quicker, faster, more economical for clients.”

Unintended consequences

Tethering transformation services to technology solutions might deliver growth, but it’s not without its own risks. By presenting themselves as providers of third-party gen AI applications, agencies might lose their distinctiveness. Isn’t one company that can hook you up with the latest OpenAI application as good as the next? And how can agencies improve those services if they must wait on companies such as Google and Adobe to develop them?

Naeger concedes that Merkle, and its rivals, are “dependent and reliant on the advancements of third-party tools.” The aforementioned partnerships, he says, give Merkle an edge via early access to new developments from Salesforce and Microsoft – though rival holding companies would each make the same argument.

TwentyfirstcenturyBrand co-founder Neil Barrie argues that agencies must strike a balance between the capabilities of their staff and those of the tools they can connect clients with. The boutique consultancy recently began offering transformation services to clients in addition to its expertise in brand strategy. Barrie notes that despite the “uncertain market,” CMOs are faced with a number of challenges that may drive demand.

“The majority of global CMOs are tasked with transforming while performing at the same time. And they’ve got constant internal flux, AI, emerging tech, environmental and cultural challenges.”

But he cautions: “Tech can open up better value for companies... but you have to make sure you bring the people with you. We’re not in any way Luddite – we’re all about using tech to get to better outcomes – it’s just about getting the balance right.

“When you’re talking about technology, you’re also talking about behavior change. You need those two things together. There is a 78% failure rate for change initiatives. That’s because a lot of the time, the human side is disproportionately neglected.

Accenture’s Temple adds: ”Everyone is focused on what AI can do. It’s equally important to focus in on what it can’t do.”

He’s confident that Accenture Song can retain its premium by ringfencing the parts of its offering that require creative thought that yield ideas that “make your hair stand on end.” Furthermore, he adds, the speed enabled by AI tools will be so valuable that clients will want to pay for it. “We still retain that premium service because we get the time back,” he says.

Naeger suggests that, despite the use cases developed by Merkle and its competitors, it’ll be a while before gen AI applications provoke further demand for transformation services. “Marketing budgets are planned annually, and right now clients need to shorten that planning cycle. You can’t wait for the third quarter, you need budget flexibility,” he argues.

In the meantime, he suggests agencies should be focusing on proving the effectiveness of an AI-enabled transformation. “We are seeing increased demand for understanding the outcomes of what AI is producing. We’re seeing a lot of decisions made today that do the bright and shiny thing versus a business need. [The business need] has been our larger focus.”

At Spark Foundry, Angelides also acknowledges that shifting the transformation offering to a tech-enabled one will have implications for the company’s business model. Like Naeger, his thoughts also turn towards effectiveness.

“It sounds like such a bloody cliché but the biggest differentiator will continue to be people,” he says. As such, the agency can retain its value proposition by having its people test and prove the worth of any given tech solution, “having a culture of confidence and expertise.”

“We call ourselves the acceleration agency. But acceleration doesn’t just mean speed. Speed is great if you’re heading in the right direction… but it doesn’t matter how quickly you’re going if it’s not effective.”

Agencies Agency Models Digital Transformation

More from Agencies

View all


Industry insights

View all
Add your own content +